Advice on this design, please!

Evening,

This will be my first build and I've been doing lots of research into the design of it based on what I would like to do with it (milling aluminum and other soft metals). I am still very uneducated in the world of DIY CNC machines, so please bear with me. I originally planned on doing affixed gantry/moving table design, but as I was designing it on Fusion 360 I realized my spindle was getting farther and farther out from my gantry (given the constraints of the components I plan to buy. Chinese ball screws and linear rails). So I decided to move the Z-axis to the main upright supports and allow the entire gantry to move up and down (as shown in the pictures attached. The design isn't finished. I have not designed the moving table yet). To me this seems like a good idea and a very rigid build....but what do I know! This is all new to me. One of the issues I can see is the gantry will be heavy. It's going to be 8"x4"x0.5" rectangular steel tubing, in fact, the entire machine will be steel. Since this gantry will be so heavy will I need to invest in much more powerful motors to move the thing up and down? Are there any major issues that you can see with this design? Any tips on repositioning the ball screws or linear rails? Just keep in mind this was a quick sketch up of the design idea and is by no means a finished product.

Any criticism would be greatly appreciated. I would like to get started on ordering parts.

Cheers!

Curt

Attachment 483167
Attachment 483169
Attachment 483171

Hi Curt - Your quite a away away from ordering parts. What about the rest of the machine? In principle the configuration is good. There are many machines that use a lifting gantry so no precedent set there. Yes the gantry can be heavy and in a moving column design the gantry just translates so no lifting there. The Z axis is the only lift needed. A 5mm screw will lift 100s of kg so no need to worry about that. So keep detailing out the machine. Your column size and attachment to the table rails will be critical to maintain a stiff design. Looks like you will weld the machine. Welding will distort all your parts and you need to think about stress relieving the parts before you final machine. You seem to have only one bearing on the column rail. You will need two bearings per rail for geometric stability. The columns only need one rail, two rails not needed. That's a start. Peter

Hey Peter,

Thanks for the advice! Haha, yeah I have the rest of the machine finished in a previous design (I think this is design #6 now). I just need to stick this gantry design onto my moving table and with a few tweeks here and there I should have a complete design. Or at least a starting point. I'll post the completed design as soon as it's finished. Ok that's great to hear about the weight of the gantry not being an issue. I planned on using beefier linear rails (maybe 25mm) for the Z axis as well as a beefier ball screw. You think a 5mm pitch ball screw for the z axis would be a good idea in this situation? I was planning on using 2010 ball screws for the y and z axis and 1610 for the x axis. But with my placement of the ball screws I really have no room restraints so I could just go 2010 or 2005's all around. Originally I wanted to weld the machine, but after reading about heat treatment to relieve the stress caused by welding, I'm leaning more towards bolting as much as I can together. All the joining faces and mating surfaces for the linear rails and ball screw supports will be machined flat as well. Oh ok, so you think two rails on either side of the columns will be overkill? That's good to know. I'll cut it down to one rail for each side with two bearing blocks per rail then.

I really appreciate the advice!

Curt

Hi Curt - Only #6 !! 6 more to go I think plus 4+ rabbit holes... If your into maths the Hiwin site publishes the formulas for sizing screws. Do not go bigger then required because the rotational inertia of the screw is significant and the motor will end up just turning the screw vs doing the job. What is the length of the screws? Being a mill 5mm for all axis maybe good. Depends on the speed you want to cut at. I'd consider servos they are a better match to what you want to do then steppers. After 100s of designs and 1000s of hours I have settled on this configuration for my mill. Being a first machine and a mill you have large hurdles to jump. But its good to aim high. Depending on the size of the jobs you want to do will determine various machine elements. So
1) a lifting gantry design with moving gantry has extra degrees of freedom that need to be controlled. If not built really stiff it will vibrate easily and may not be worth the concern of the extra overhang you mentioned
2) if its a lifting gantry with moving bed that's easier to do
3) depending on what accuracy you are aiming at and what surface finish you are expecting to achieve will determine the quality of the motion parts required. Its easy to spend your entire envisaged budget on a single high quality screw, I'm sure... machining centres cost a lot of money for a reason...
Peter

Mills need to be considerably stiffer then routers and they need fine control over their axis. That's why they try to stay with 3 screws only. Getting your LHS and RHS to match Z will be difficult using cheap screws. But this comes down to what accuracy you need in the machine. Your machine has 5 screws with no positional feedback. That's a mission....

You say moving table so you have 4 screws thats a bit easier... A good mill always has the tool forces within the footprint of the bearings so no secondary moments are created eg the Mori. Your mill will always have secondary moments to deal with that create vibration and other issues. But this comes down to the envelope you are trying to achieve..

Hi Curt - To explain the accuracy a little better. You have two columns under control from ballscrews. Generally if you buy on-line quality screws they will be grade C7 this means at best they will have an accuracy of 0.050mm per 300mm of travel ie they maybe at best be within 0.05mm of the correct position. The longer the travel the worst this gets. Since you have two screws one could be high the other low so the gantry could be 0.1mm out of level. Then there is backlash (a C7 screw has an axial play of 0.050mm) and the electronic tolerances issues to add to the mix. If you are expecting to get 0.01mm accuracy then you have to rethink the system. Peter

https://www.hiwin.us/wp-content/uplo...ew-Catalog.pdf

Being a mill I would place a bridge across the top of the columns and I would diagonally brace the columns back to the machine bed. This is not for primary stiffness its for vibration mitigation. Cantilevers have a habit of vibrating... The bridge is to help the two columns co-operate under torsion loads and to stop vibration.

Hey Peter,

Thanks again, I took your advice and made some changes to the current design. I'll list what I decided to go with:

Y-Axis - 1200mm long 20mm linear rails and 1000mm 2005 ball screw with double anti backlash nuts (or 2010 ball screw)
X-Axis- 700mm long 20mm linear rails and 800mm 1605 ball screw with double anti backlash nuts (or 1610 ball screw)
Z-Axis- 600mm long 20mm linear rails and 600mm 1605 ball screws with double anti backlash nuts (x2) (or 1610 ball screws)

Still deciding on the motors. You mentioned servos over stepper motors so I'm just looking into that now to see if the budget will allow for it at this time.

As for the structure:

-table made from 3"x3"x0.375" mild steel square tubing (I'll have to figure out a design for the legs)
-columns made from 6"x3"x0.375" mild steel rectangular tubing
-gantry made from 8"x4"x0.5" mild steel rectangular tubing
-moving table made from 0.5" steel plate and roughly 21.5" x 24"

I'm going to try and avoid welding as I have no way of heat treating the metal afterwards. I'll bolt as much as I can. I took your advice and added a brace above the columns. I went with a threaded rod with double nuts on either end for added rigidity and (at least I hope it will help). I thought maybe that would help with getting the columns perfectly parallel with each other.

The useful cutting window will be roughly 21.5" x 24" x 10". I know my Z axis travel seems large. I liked the idea of having the extra room, I just hope accuracy doesn't suffer too much with it. I would much rather accuracy over speed, hence my leaning towards the use of 5mm ball screws.

I'm torn between continuing with this design (which I like the most) or going back to a fully rigid gantry. Do you thing having the double ball screws on the Z-axis is going to be a problem with accuracy?

Any other tips would be greatly appreciated!

Cheers,

Curt

Hi Curt - The threaded rod is not substantial enough it needs to be a similar size to the columns. My commentaries context is the comment "machining aluminium and soft metals". The braces need to be from the top of the columns back to the base frame. The plate webs are good but not good enough to stop a cantilever vibrating. You will need 10" Z (or more) by the time you use a vice and a part and a tool 10" is not much space. Your bed will need to be much thicker. Plunging loads will be in the order of 200kg plus and that bed plate will bend under such a load. I suggest you add a third rail in there at the centre...

Servos may seem expensive but these are the heart of the machine. Its worth the $$$ and a fine screw gives you lots of force but not much speed. A servo gives you 3000rpm (3000x5=15000mm/min good speed you will need this to cut Al with a high speed spindle) look up chip loads and chip thicknesses, tool specs and you will understand that better... and a stepper will top out at 700rpm with no torque left.

Accuracy is something you have to decide for your application. If you are to make watch parts forget it... if its motorcycle accessories its Ok. What do you want to do with the mill? Peter

this sort of space frame may give you ideas, especially a bolted up frame. If your bolting consider Channels as then you can bolt up hard. Channels are no good in torsion but in bending they are good... if you have oxy or propane oxy you can torch braze. This will give you zero warpage and is easy to do... then you can build as you have drawn. A very good process is to get someone with TIG to tack it togther of if you have oxy you can oxy tack it, then once tacked and checked for square etc braze it together. I use tobin bronze.

Hi Curt- Looking at your design again:
1) You have mounted the column rails in the centre of the section. This area is quite flexible. They need to be on an edge of a section vs in the middle
2) You need to take full advantage of the footprint so some rear columns connected to the main columns, cross braced with a bridge on top of both would be really good. Something like the cubefoot. Every possible space that you can used must be used to create stiffness. If you look at a commercial VMC of the same foot print it weighs a couple of tonnes minimum for a reason
3) speak to your machinist about how to register the top and bottom rail lands on the gantry. Look at the rail manufacturers specifications and check the machinist can do this. They are placed on one side usually so it can be done in one set up. You maybe able to do it in one depends on the tooling and machine the machinist has
4) The rectangular frame where the columns sit need to be triangulated fwd and rearward to the longitudinal beams. This will improve its bending and torsion stiffness
5) Find out the vice size your likely to use and check 10" is enough Z probably not
6) Your saddle needs side webs or it needs to be much thicker something like 30mm plus. Side webs are more efficient and provide more inertia
7) Start working on the bolted connections they will require a bit of thought

Peter

Hi Curt– There are two main structural philosophies 1) monocoque & 2) space frame. Your machine base is a frame your columns are monocoque for instance. Your columns deflection will be governed by this eqn def=(loadxlength^3)/(3xmodulusxinertia) length and inertia dominate the issue.


If you built a space frame it could be lighter and stiffer. A space frame uses the frames geometry for stiffness vs its sectional inertia. A triangular cell or frame is very stiff even if its bolted together. A rectangular cell depends on its corner stiffness for its global stiffness so we try to triangulate as much as possible.


Plus cantilevers, like tuning forks vibrate and vibration is something machine designers try to minimise.


Back to frames. A frame can only resist forces in its plane of construction. If a force is applied outside the cells plane that has to be resisted by a moment. So we try to keep forces arranged so they are in the frames plane. Large moments are difficult to deal with. A machine is better (more rigid) if it is dealing with direct forces only.


The cutting tool is responsible for the machines main forces so in the transverse direction it would be good to have the bearings in the tools plane. But this is difficult. But this does mean its better to have the bearings on the front of the column and try to get them as close to the tool plane as possible. Maybe the gantry can be set back a little like the canted columns on a moving column machine. Your bearings inside the column are trying to do this as well. But inside will present manufacturing issues that you will need to work through.


Then there's longitudinal requirements, again a triangular truss is good. .Then there is the rule to take advantage of all space that does not interfere with function or footprint. This results in a triangulated tunnel or portal frame over the rear of the machine. Since your using F360 you can do a quick FEA on your design and a portal to see the difference.


Since its triangulated the tubes don't have to be big as its the cross sectional area that provides the stiffness not the sectional inertia. This occurs as we are not bending something, again bending is moments which we are trying to remove... see attached image cantilever vs space frame.


Hope that helps. There is a space frame mill in here somewhere... Peter

Hi Curt - I'm going through this exercise for my mill at the moment so I'll copy it in here:

https://www.guhring.com/ProductsServ...omResults=True

A typical chip thickness for aluminium is 0.08mm see chart attached... this is for an 4F tool. Feed=Nxchip t x RPM so if you have a 20k spindle you want to feed at 6400mm/min. If you use 5mm pitch thats 1280rpm so no good unless its a servo as they top at 3000rpm. If 10mm pitch then thats 640rpm and a stepper is well on its way to less then half its torque plus half its torque is taken up turning ballscrews if they are accelerating... So that's why I suggest servos. If you cut your spindle back to 10,000rpm you then halve these but you are running the spindle at an inefficient speed.... So to get to a sweet spot with steppers you will use 1F tools. so your feed at 20k for a 1F tool is 0.08*20000= 1600mm/min and at 5mm pitch thats 320rpm so within a steppers high torque band. A 2F tool takes it to the limit at say 640rpm. Seriously look at servos much more suited to the task... Motors are the heart of the machine, structure is the soul gota get both right... Peter

Hi Curt - sorry to overload you but I'm working thru a similar design but smaller so thought I'd share stuff with you. I'm new to F360 so I'm struggling with a few things but I ran a column today thru generative and heres the results. The result is a basic tapered I beam. If the load is at an angle it skews the I beam to the same plane as the load. I beams are poor in torsion (applied moments) but if you make the space frame I discussed the frame takes care of the torsion and the open sections I's or C's take care of the local loads. Many of these columns are what's called short beams and shear transfer is a big internal condition. Shear deflection is more dominant in short beams then global deflection. Thats why the optimiser has left the web down the middle vs making it hollow.

eg you placed a rail in the middle of a hollow side so there is no metal behind the rail to transfer the shear load. It has to deflect in what's called membrane action and strain stiffen before it can transfer load. Like a hammock. Its floppy until you get into it and stiffen it via membrane action. Local deflection is something you do not want.

Hollow sections are optimised for long beams with distributed loads not point loads. CNC have moving point loads so tapering sections are not easy to meet the structural issues. So just some food for thought if you want to go down an "unconventional route". I beams and Cs are easier to bolt things together with or construct from flat plate. You just need to spot machine the faying surfaces so they are flat and in the desired spots. Peter

Hey Peter,

I finally got some free time to update the design. Thank you so much for all the info! That's super helpful. I never even thought about brazing, but that actually makes a lot of sense. I've gone down a bit of a rabbit hole looking into brazing and it seems like a good option. I have a MIG welder I could maybe do small tacks with then braze the rest. I did this design before reading your comments on using an I-beam or C-beam for the columns and I actually really like that idea. I think I'll try to incorporate that. I've put NEMA 23 motors on this design only because I haven't found proper servos to use. Do you have any suggestions? That pdf you posted is very helpful, it spells it out nicely that a stepper motor might not be up to the job. I liked your comments about putting the Z-axis bearings more in the same plane with the spindle then setting the gantry back. That won't work with this current design, but that's something I'll keep in mind.

Again, I appreciate all the advice it's really helping me finalize this design.

Curt

Morning Curt - I generally get my motion parts from BST Automation. BST only have C7 grade ballscrews and for a mill I suggest you get a better grade. They have good servos, at least start there. Your braces can be smaller currently they have very long connections called mitres and these will be tedious brazing. Make them square. They can also be a size down in width so the connection is gapless. If you use sections of same size the radiused edges produce large gaps. If welded or brazed these gaps create shrinkage and distortion. You need to add the section radius in the model as its significant and by adding this you won't place something too close to the edge and you will see gaps that will affect your design... Keep at it. Its time to think about how you are going to mill the rail foundations and any other spots. The current design you can't mill the column lands after brazing. Peter

https://bstmotion.aliexpress.com/store/314742

Hi Curt - Regarding Brazing : There are two styles 1) braze welding and 2) flow brazing 1) You use an oxy acetylene or oxy propane torch with a hot cone. Its the same as TIG welding, you heat up a small amount of metal to red and drop a bead onto it then move along to the next bead etc. This is a well controlled process 2) flow brazing you use a big tip or a rose and heat a large area of metal and when braze is introduced it will flow along the entire area heated up. Both can be used here. I recommend not to use flux coated rod I find it to be painful to use. Clean all joints up to 50mm away from the joint so it is shiny metal. Apply a flux slurry (flux plus water until thick a slight flow but not runny) to all the joints about to be brazed. You can tack braze one side then go to the opposite side. If you MIG use a big preflow time so you minimise charring up your joint. A scotchbrite small belt on a hand linisher is great for cleaning up. If you are really into brazing and you have a good welding supplies at hand ask about in-line fluxing its the bees knees. In your flux bottle have a 400mm long piece of stainless steel rod say 2 or 3mm diameter for placing flux (stainless so it does not rust and muck up your flux). If flow brazing you can wipe the flux coated SS rod along the joint after brazing and it will smooth and clean it up nicely before it cools enough to freeze.

Now as per welding don't be tempted to join all of one joint together at the same time. Join one side then the opposite side and move along. Come back later to close out the joint. Move around the structure and don't concentrate on one area too much. Good luck... get some steel and have a practice....You need to learn to read the flux and the metal colour a nice rosy red and clear flux with clean steel and its smooth brazing. Peter